Advances in Chemical Engineering and Science, 2012, 2, 474-480

doi:10.4236/aces.2012.24058 Published Online October 2012 (http://www.SciRP.org/journal/aces)
I nvestigations into Optimization Models of Crude Oil
Distillation Column in the Context of Feed Stock
and Market Value
Lekan T. Popoola
1
, J amiu A. Adeniran
2
, Solomon O. Akinola
3

1
Department of Petroleum and Chemical Engineering, Afe Babalola University, Ado-Ekiti, Nigeria
2
Department of Chemical Engineering, Ladoke Akintola University of Technology, Ogbomoso, Nigeria
3
Department of Computer Engineering, Afe Babalola University, Ado-Ekiti, Nigeria
Email: popoolalekantaofeek@yahoo.com

Received July 14, 2012; revised August 25, 2012; accepted September 3, 2012
ABSTRACT
This paper proposes optimization models of crude oil distillation column for both limited and unlimited feed stock and
market value of known products prices. The feed to the crude distillation column was assumed to be crude oil contain-
ing naphtha gas, kerosene, petrol and diesel as the light-light key, light key, heavy key and heavy-heavy key respec-
tively. The models determined maximum concentrations of heavy key in the distillate
( )
D
o
HK and light key in the
bottom
(
B
o
LK
)
for limited feed stock and market condition. Both were impurities in their respective positions of the
column. The limiting constraints were sales specification concentration of light key in the distillate [ ( ) ( )
D D
ss
LK LK > ],
heavy key in the bottom [ ( ) ( )
B B
ss
HK HK > ] and an operating loading constraint of flooding above the feed tray. For
unlimited feed stock and market condition, the optimization models determined the optimum separation [
( )
D
o
HK and
(
B
o
LK
)
] and feed flow rate (
o
F )that would give maximum profit with minimum purity sales specification constraints
of light key in the distillate and heavy key in the bottom as stated above. The feed loading was limited by the reboiler
capacity. However, there is need to simulate the optimization models for an existing crude oil distillation column of a
refinery in order to validate the models.

Keywords: Feed Stock; Crude Oil Distillation Column; Constraints; Market Condition; Reboiler; Crude Composition;
Optimization
1. I ntroduction
While the use of distillation dates back in recorded his-
tory to about 50 B.C., the first truly industrial exploita-
tion of this separation process did not occur until the 12th
century when it was used in the production of alcoholic
beverages [1]. Crude oil distillation is the separation of
the hydrocarbons in crude oil into fractions based on
their boiling points which lie within a specified range [2].
The separation is done in a large tower (Crude Oil Dis-
tillation Column) that is operated at atmospheric pressure.
The tower contains a number of trays where hydrocarbon
gases and liquids interact. The liquids flow down the
tower and the gases up [3]. The fractions that rise highest
in the column before condensing are called light fractions,
and those that condense on the lowest trays are called
heavy fractions [4]. The portion of the column above the
feed iscalled the rectifying section and below the feed,
the stripping section.Globally, more than 80 million bar-
rels of crude oil are refined daily. In the US, 146 refin-
eries operate, employing over 65,000 people and pro-
ducing a total value that exceeds $151 billion [5]. Figure
1 gives a schematic representation of the process over-
view of a crude oil distillation column.
Crude Distillation Units (CDUs) are key process plants
in a petroleum refinery as changes in these units have a
great impact on product yield and quality. It is recom-
mended to operate these units at optimal conditions from
technical and economical points of view. Mathematical
modelling has become very common to develop these
optimization studies [6]. Many conventional ways of
optimizing crude oil distillation column such as total
Copyright © 2012 SciRes. ACES
L. T. POPOOLA ET AL. 475


Figure 1. Process overview of crude oil distillation column.

energy requirement, heat integration, enhanced crude
blending, crude oil energy consumption, operating condi-
tions, products specification and many more had been
discovered by previous researchers. Yu et al. [7] pro-
posed an on-line soft-sensor for the quality control and
energy optimization of crude distillation column. A
two-stage soft sensor was developed for the estimation of
internal liquid and vapour loads, and product quality
specifications of side streams. The measurements of the
soft-sensor were used to derive the constraints of the en-
ergy optimization system as well as feedback to the qual-
ity control system. Gadalla et al. [8] developed optimiza-
tion framework which included shortcut models for the
simulation of the existing distillation column and a retro-
fit shortcut model for the heat exchanger network. The
process was optimized by changing key operating pa-
rameters while simultaneously accounting for hydraulic
limitations and design including the performance of the
existing heat exchanger network. A reduction in energy
consumption and operating costs of over 25% was
achieved. In 2003, Okeke et al. [9] proposed the design
and optimization of a crude distillation unitat the Nigeria
Port Harcourt refinery in the context of total energy re-
quirement. The modelling procedure used was a stage-
wise approach to determine opportunity for optimization
using sequential quadratic programming (SQP). They
concluded that the programming technique was ideal for
solution of the algorithm achieving quick convergence
and 8% gasoline yield increment. Domijan et al. [10]
made improvement on research done by Yu et al. [7] and
Okeke et al. [9] by considering off-line energy optimiza-
tion model for crude distillation unit. The CDU model
was a non-stage-by-stage, steady-state model and cor-
rected by real process values. The model calculated the
yields and properties of the products based on feed in-
formation and product specifications. The optimal solu-
tion obtained showed that the energy consumption could
be decreased 3.2% over the normal operating condition.
Fazlali et al. [11] also considered an energy saving op-
tion in refinery industry but used operating conditions to
optimize the distillation columns. The atmospheric dis-
tillation unit of Iran-Arak-Shazand petroleum refinery
was subjected to optimization efforts using a simulator
with theaim to earning more overhead products. Results
demonstrated that changes in the real operating condi-
tions increased the overhead products with desirable
quality. Kozarev et al. [12] investigated the optimization
of crude distillation units using the knowledge of com-
puter aided steady state control to optimize the mode of
operation of a crude oil column with respect to the prod-
uct prices, energy consumption and standard require-
ments of products compositions. A simple non-iterative
Copyright © 2012 SciRes. ACES
L. T. POPOOLA ET AL. 476
mathematical model of a crude oil distillation tower to-
gether with an appropriate technique was presented. The
results revealed that a good separation at considerably
low quantity of reflux could be realized. The use of the
short-cut model made the optimization procedure to be
very fast. Benyoucef [13] worked on oil refining plan-
ning under price and demand uncertainties using Algeria
as a case study. His paper aimed at analysing the Alge-
rian refining industry development in the presence of
uncertainties for both domestic products demand and the
international markets. Considering multiple uncertainties
on demand and oil price, the model provided the produc-
tion levels, the rate units running and the foreign trade of
products by the year 2030. Hassan et al. [14] improved
on the work done by Benyoucef [13] by using linear pro-
gramming modelling to improve oil refinery productivity
but through enhanced crude blending. The objective of
their study was to develop a mathematical programming
model for solving a blending problem in a major refinery
Alexandria, Egypt with the objective of maximizing
naphtha productivity. Their study showed that the devel-
oped linear programming model for the blending prob-
lem yielded better overall naphtha productivity for the oil
refinery studied as compared to the results obtained from
the commercial software.
2. Proposed Optimization Models
The optimization models proposed are applicable for
known product prices of crude oil distillation column in
order to obtain the specified separation for the least oper-
ating cost under the conditions of both limited and
unlimited market and feed stock. It is impossible to
maximize the profit without knowing the prices of ter-
minal products of the distillation column and taking into
account all other aspects of the overall plant in which the
column is a part. The feed to the crude distillation col-
umn is assumed to be crude oil containing naphtha gas,
kerosene, petrol and diesel as the light-light key, light
key, heavy key and heavy-heavy key respectively. The
limited market and feed stock optimization models are
presented for the conditions in which the top products are
more valuable than the bottom products and vice versa.
2.1. Product Prices Known with Limited
Market and Feed Stock
Condition 1: Top products are more valuable than the
bottom products
Since it is assumed that the top products have the
highest unit prices, they must be produced at minimum
specified purity such that the heavy key component in the
top product (HK
D
) is the impurity present in the distillate
and hence, the control component. The maximum con-
centration of the heavy key in the distillate is calculated
thus (see below):
where
( )
D
o
HK = Optimum concentration of heavy key in the
distillate, mol/m
3
,
LK
F
= Concentration of the light key in the feed,
mol/m
3
,
B LK = Concentration of the light key in the bottom,
mol/m
3
,
LLK
F
= Concentration of the light-light key in the feed,
mol/m
3
,
(LK
D
)
ss
= Sales specification concentration of light key
in the distillate, mol/m
3
.
When the change in operating cost equals the change
in product worth,
( )
( )
( )
( )
( )
i
i
B B
L
D
PD PB CL
LK LK
(
c
c
(
(
÷ =
(
(
c c
(
¸ ¸ ¸ ¸
(2)
where
PD = Price of the distillate, ₦/sec,
PB = Price of the bottom, ₦/sec,
D = Distillate concentration, mol/m
3
,
CL
i
= is the unit cost of operation per unit of internal
reflux, ₦(s/m
3
),
L
i
= internal reflux, m
3
/s,
B LK = Concentration of the light key in the bottom,
mol/m
3
.
The simultaneous solution of Equations (1) and (2) gives
the the optimum values of HK
D
and B LK . The light key
in the distillate (LK
D
) is the purity component in the top
product and thus, its purity concentration must not be
less than that of the sales specification. Hence, the limit-
ing constraint equation for Equations (1) and (2) above is:
( ) ( )
D D
ss
LK LK > (3)
where (LK
D
)
ss
is the sales specification concentration of
the light key in the distillate, mol/m
3
.
Condition 2: Bottom products are more valuable than
the top products
When the bottom products are assumed to be more
valuable than the top products, light key in the bottom
(LK
B
) becomes the impurity and its maximum concentra-
tion in the bottom products should be determined as the
ontrolling component. Thus, we have c

( )
( ) ( ) ( ) ( ) ( ) ( ) ( )
( )
( )
1 [ B B
F F D F F
ss
D
o
B
F
LK LK LLK LK LLK LK LK
HK
LK LK
+ ÷ ÷ + ÷ (
¸ ¸
=
÷
(1)
Copyright © 2012 SciRes. ACES
L. T. POPOOLA ET AL. 477
( )
( )
( ) ( ) ( )
( )
( ) ( )
( )
( )
( )
max
1 1 D B
F F F F F
o ss
B
D
F
o
HK LLK LK HK HK LLK LK HK
LK
LK HK
( (
÷ + + + ÷ + ÷ (
¸ ¸
¸ ¸ ¸ ¸ ¸
=
÷
B
ss
(
¸
(4)

where
(
D
o
HK
)
= Optimum concentration of heavy key in
the distillate, mol/m
3
,
( )
B
ss
HK = Sales specification concentrationof heavy
key in the bottom, mol/m
3
,

( )
F
HK = Concentration of heavy key in the feed,
mol/m
3
,
( )
max
B LK = Maximum concentration of the light key
in the bottom, mol/m
3
.
Other variables are as stated for Equations (1) and (2)
above.
The heavy key in the bottom
( )
B HK is the purity
component in the bottom product whose concentration
must not be less than that of the sales specification.
Hence,
( ) ( )
B B
ss
HK HK > (5)
The column is assumed to approach an operating con-
straint of flooding above the feed tray (because of exces-
sive entrainment of liquid in the vapour) as
( )
D
o

determined from Equations (1) and (2) above is lowered.
The loading equation for this constraint is given as:
HK
( )
( )
1 3 2
1 1
i
L B
a a P F a
F F
(
+ + ÷ + ÷ =
(
¸ ¸
0 (6)
where
a
1
, a
2
and a
3
= Coefficients of experimental loading
equation,
P = Column pressure, Pa,
F = Feed flow rate, m
3
/s,
L = L
i
= Liquid flow rate in column at the point at
which flooding occurs, m
3
/s,
L
i
/F and B/F are obtained from the operating control
equations for reflux and bottom product flow rate.
B = Bottom flow rate, m
3
/s.
As
(
D
o
HK
)
is reduced from optimum, L
i
/F and B/F
change. The value of
(
D
o
HK
)
will be obtained at a
point where Equation 6 is equal to zero. That value of
(
D
o
HK
)
is used in Equation 4 to obtain
( )
max
B LK . The
flow chart algorithm for limited market and feed stock is
shown in Figure 2.
2.2. Product Prices Known with Unlimited
Market and Feed Stock
For an unlimited market, the feed flow rate and separa-
tion resulting in maximum profit rate for the existing
products must be determined. Operation for a column
under this condition will always be against an operating
constraint because of the unlimited market and feedstock.
The overall optimization problem involves determining
the optimum separation [i.e.
(
D
o
HK
)
and
( )
B
o
LK ],
feed flow rate (i.e.
o
F ) and column pressure [
( )
o
P ] that
will give maximum profit. The optimum concentrations
of the separation,
( )
D
o
HK and
(
B
o
LK
)
, are calculated
using Equations (1) and (4) above but with the following
minimum purity constraints:
( ) ( )
, D D B B
ss s
LK LK HK HK = =
s
(7)
( ) ( ) ( ) ( )
max max
, D D
B B
o o
LK LK HK HK = = (8)
The maximum feed flow rate is calculated by using:
( )
( )
max 1 2 1 2
2
3 1 2

c
c
F a a d d P T
a d d P T B L
(
= + + ÷
¸ ¸
(
+ + ÷ + ÷
¸ ¸
(9)
where
a
1
, a
2
and a
3
= coefficients for condenser-loading
equation that are determined by column tests
d
1
and d
2
= coefficients determined off-line from cor-
relation of data obtained from flash calculations at the
average composition of the existing overhead vapour.
P = Column pressure, Pa,
T
c
= Temperature of the coolant to overhead vapour
condenser, ˚C,
B = Bottom product flow rates from the output of op-
erating control equation, m
3
/s,
L = Reflux flow rates from the outputs of operating
control equation, m
3
/s,
F
max
= Maximum feed flow rate required to load the
condenser for optimum separation, m
3
/s.
For Equation 9 above, the temperature of the overhead
vapour (T
o
) is a function of column pressure only while
other independent inputs are negligible.
Assuming the feed loading is limited by the reboiler
capacity, the column must then be operated at minimum
value in order to gain maximum reboiler capacity. Hav-
ing known the optimum separation and column pressure,
the the feed rate can now be increased up to the maxi-
mum capacity of the reboiler which gives the most prof-
itable operation. The feed flow controller set point that
gives the maximum reboiler heat input rate is calculated
thus:
Copyright © 2012 SciRes. ACES
L. T. POPOOLA ET AL. 478


Figure 2. Flow chart optimization algorithm of crude oil distillation column for limited market and feed stock.

( )
( )( )
max
1
B
i
F v f
V
F
L B
K T T
F F
=
| | | |
+ + ÷ ÷
| |
\ . \ .
(10)
where
F = Feed flow controller set point, m
3
/s,
( )
B
V
max
= maximum reboiler heat input rate, J/s,
F
K = a constant equal to the specific heat of the feed
divided by the heat of vaporization,
v
T = Temperature of vapour above the feed tray, ˚C,
f
T = Temperature of the feed at column entry, ˚C,
i
L
and
B
F
are obtained from the operating control
F
equations used to achieve a suboptimum operation (Fig-
ure 3).
3. Discussion
The existences of the feed stock and market situation
which may be limited or unlimited are among major fac-
tors in the determination of process optimization of crude
oil distillation column. The proposed modelling equa-
tions are applicable for the process optimization of crude
oil distillation column with products purity specification.
Hence, the products prices and their purity specifications
from customer need be known for this goal to be
achieved. For known products prices of crude distillation
column with limited market and feed stock, the simulta-
neous solution of Equations (1) and (2) gives the optimum
values of the heavy key in the distillate (petrol) and light
key in the bottom(kerosene). Assuming the top products
are more valuable than the bottom, then the light key in
the distillate (kerosene) is the purity component in the
top product and thus, its purity concentration must not be
less than that of the sales specification. Equation 3 is then
the limiting constraints for the simultaneous solution of
Equations (1) and (2). When the bottom products are as-
sumed to be more valuable than the top products under
the same condition of limited market and feed stock, the
light key (kerosene) becomes the impurity in the bottom
Copyright © 2012 SciRes. ACES
L. T. POPOOLA ET AL. 479

Figure 3. Flow chart optimization algorithm of crude oil
distillation column for unlimited market and feed stock.

and thus, its maximum concentration is calculated by
using Equation (4). However, the heavy key (petrol) in
the bottom becomes its limiting constraint whose con-
centration must not be less than that of the sales specifi-
cation (Equation (5)). To get the value of
(
D
)
HK
o
to be
used in Equation (4), the column is assumed to approach
an operating constraint of flooding above the feed tray.
The loading equation for this constraint is stated as Equa-
tion (6). As the value of
(
D
)
HK
o
gotten from Equation
(1) is reduced from optimum, the values of L
i
/F and B/F
change. Thus, the value of
(
D
o
HK
)
that satisfies Equa-
tion (6) is obtained to solve for
( )
max
B LK in Equation
(4) while other variables are stated. When unlimited
market and feed stock exist with known product prices,
the optimum separation [i.e.
( )
D
o
HK and
(
B
o
LK
)
] is
determined using Equations (1) and (4) but with the lim-
iting purity constraints stated in Equations (7) and (8).
The maximum feed flow rate that gives maximum profit
rate is determined using Equation (9) in which the tem-
perature of the overhead vapour (T
o
) is a function of the
column pressure. If the feed loading is assumed to be
limited by the reboiler capacity instead of the column
pressure, the feed flow controller set point that gives the
maximum reboiler heat input rate is evaluated using
Equation (10). The proposed optimization models can be
used to simulate an existing crude oil distillation column
of a refinery.
4. Conclusion and Recommendation
The optimization of crude oil distillation column in the
context of feed stock and market condition can be
achieved using the stated models. The products prices
from the crude oil distillation column must be known and
the consumers’ products specification were put into con-
sideration using the products limiting purity constraints.
Also, the effect of operating cost on the products worth
was considered by assuming that the change in operating
cost equals the change in product worth. Nevertheless,
the proposed optimization modelling equations are ap-
plicable for both the conditions of limited and unlimited
feed stock and market situations for known product
prices. The situation in which either the top products or
bottom products being valuable than each other for the
condition of limited feed stock and market condition had
been examined. However, it is recommended that the
point of introducing the feed into the column be known
to validate the assumption of flooding above the feed tray.
Finally, the proposed models should be tested on an ex-
isting crude oil distillation column of a refinery for future
research in order to validate the applicability.
REFERENCES
[1] J. G. Kunesh, T. P. Ognisty, M. Sagata and G. X. Chen,
AIChE Spring National Meeting, Atlanta, 1994.
[2] K. H. Bawazeer, “Prediction of Crude Oil Product Qual-
ity Parameters Using Neural Networks,” M.S. Thesis,
Florida Atlantic University, Boca Raton, 1996.
[3] W. Heather, “Refining Crude Oil,” The New Zealand
Refining Company Ltd., Whangarei, 2003.
[4] Z. E. H. Tonnang, “Distillation Column Control Using
Artificial Neural Networks,” M.Sc. Thesis, University of
Ibadan, Ibadan, 2003.
[5] B. Liptak, “Distillation Control and Optimization,” 2009.
www.controlglobal.com
[6] D. C. López, C. A. Mahecha, L. J. Hoyos, L. Acevedo
and J. F. Villamizar, “Optimization Model of a System of
Crude Oil Distillation Units with Heat Integration and
Metamodeling,” CT & F—Ciencia, Tecnología y Futuro,
Vol. 3, No. 5, 2009, pp. 159-173.
[7] J. J. Yu, C. H. Zhou, S. Tan and C. C. Hang, “An On-
Line Soft-Sensor for Control and Optimization of Crude
Distillation Column,” Research Institute of Industrial Pro-
cess Control, Hangzhou, 1997.
[8] M. Gadalla, M. Jobson and M. Smith, “Optimisation of
Existing Heat-Integrated Refinery Distillation Systems,”
Ph.D. Thesis, University of Manchester Institute of Sci-
ence and Technology (UMIST), Manchester, 2002.
[9] E. O. Okeke and A. A. Osakwe-Akofe, “Optimization of
Copyright © 2012 SciRes. ACES
L. T. POPOOLA ET AL.
Copyright © 2012 SciRes. ACES
480
a Refinery Crude Distillation Unit in the Context of Total
Energy Requirement,” NNPC R&D Division, Port Har-
court, 2003.
[10] P. Domijan and D. Kalpić, “Off-Line Energy Optimiza-
tion Model for Crude Distillation Unit,” Ph.D. Thesis,
University of Zagreb, Croatia, 2005.
[11] A. Fazlali, S. Hosseini, B. Yasini and A. Moghadassi,
“Optimization of Operating Conditions of Distillation
Columns: An Energy Saving Option in Refinery Indus-
try,” Songklanakarin J ournal of Science and Technology,
Vol. 31, No. 6, 2009, pp. 661-664.
[12] N. kozarev, F. Keil and S. Stoyanov, “Computer Aided
Steady State Control of Crude Oil Distillation,” J ournal
of the University of Chemical Technology and Metallurgy,
Vol. 45, No. 4, 2010, pp. 401-406.
[13] A. Benyoucef, “Oil Refining Planning under Price and
Demand Uncertainties: Case of Algeria,” 29th USAEE/
IAEE North America Conference, Calgary, 14-16 October
2010, pp. 1-19.
[14] M. K. Hassan, A. Kandeil and A. Elkhayat, “Improving
Oil Refinery Productivity through Enhanced Crude Blend-
ing Using Linear Programming Modelling,” Asian J our-
nal of Scientific Research, Vol. 4, No. 2, 2011, pp. 95-113.
doi:10.3923/ajsr.2011.95.113

Sign up to vote on this title
UsefulNot useful